
Is anyone else finding ChatGPT’s 5.* series personality more intrusive than helpful?
While the coding logic has improved, the "anti-sycophancy" tuning has created a weirdly condescending arrogance. The model now over-explains and constantly forces connections to previous conversations that have zero relevance to the current task.
It got so bad that I actually deleted my entire conversation history.. multiple gigabytes of data.. just to stop the "context poisoning."
“❤️” if your having the same issue. Comment if you have a good way/prompt to create a better personality for these models, please 🙏🏻
English





















